Gas sorption in poly(lactic acid) and packaging materials
N.S. Oliveira
a, J. Oliveira
b, T. Gomes
b, A. Ferreira
a, J. Dorgan
c, I.M. Marrucho
a,∗aCICECO, Departamento de Qu´ımica, Universidade de Aveiro, 3810-193 Aveiro, Portugal
bDepartamento de Qu´ımica, Universidade de Aveiro, 3810-193 Aveiro, Portugal
cColorado School of Mines, Golden, CO 80401, USA Available online 23 July 2004
Abstract
The solubility of nitrogen, oxygen, carbon dioxide and water in poly(lactic acid) (PLA) was measured using a quartz crystal microbalance (QCM). For that purpose an apparatus based on the quartz crystal microbalance technique was built. This apparatus was calibrated with the system carbon dioxide in atactic polystyrene and the determined precision was of 3%. Measurements were performed up to atmospheric pressure and in the temperature region from 293.2 to 313.2 K.
A study of the solubility of carbon dioxide in PLA over a temperature range from 283.2 to 313.2 K and at pressures bellow the atmospheric will be presented. The experimental results obtained were correlated with the Flory–Huggins model with an overall AAD of 6%.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Sorption; Biodegradable polymer; Quartz crystal microbalance; Poly(lactic acid) (PLA)
1. Introduction
Numerous scientists all over the world have being study- ing polymeric materials based on non-renewable fossil resources for the packaging industry. Due to the increasing concerns about the preservation of ecological systems, it is increasingly being recognized that the use of long-lasting polymers for short-lived applications is not entirely justified.
Conventional plastics that persist in the environment due to improper disposal are a significant source of environmental pollution and a waste management problem. In industrial- ized EU nations, plastics represent approximately 20–40%
of municipal waste. The time has therefore come to think of a cooperative approach for research and development of products and strategies for degradable. The elimination of conventional plastics is of interest in the wider war against environmental pollution and replacing non-degradable poly- mers with environmentally friendly degradable plastics is of major interest to both policy makers and the plastics industry[1].
The need for the development of polymeric materials based on renewable sources has led to the development of
∗Corresponding author. Tel.:+351 234 401 406;
fax:+351 234 370 084.
E-mail address: imarrucho@dq.ua.pt (I.M. Marrucho).
new materials, such as the poly(lactic acid) (PLA), which is currently being produced from a feedstock of corn rather than petroleum. The lactide monomer from which PLA is made comes from lactic acid, that has two enantiomers,l andd. Control of the (l:d) ratio monomer content is a very important feature in PLA, since it has a large effect on ma- terial properties, including the degradation time. The future widespread availability of PLA raises interesting and impor- tant questions regarding its application. Since the packaging industry, more precisely food packaging, plays a dominant role in the short term use of cheap plastic materials, their replacement with PLA could provide a significant step to- wards a greener planet. In order to preserve adequately the quality of the food, the packaging materials have to provide efficient barriers against water vapor, atmospheric gases and volatile organic compounds (VOC), preventing food degra- dation and oxidation and preserving aromas and flavors. It is commonly known that there is mass transfer both from the food to the polymer and vice versa, and that it is responsible for contamination of food and degradation of the packaging material. The food contamination issue can be minimized with the use of PLA since the lactide monomer, from which PLA is made, comes from lactic acid, a non-toxic substance available in the human body. Thus, the determination of the permeability (solubility and diffusion) of gases, flavors and aromas in polymers is of vital importance in the application
0378-3812/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.fluid.2004.06.032
of PLA in the food packaging industry. This work is the first of a study of the solubility of atmospheric gases, aromas and flavors in PLA with an (l:d) ratio of 80:20, which exhibits especially favorable mechanical, optical and degradation properties for being use in the food packaging industry.
2. Experimental method
The solubility studies of sorbents in polymers are usually done by gravimetric techniques, like the electronic sorption balance[2]or a quartz spring balance[3]or by inverse gas chromatography [4,5], although other methods like vapor pressure[6]and light scattering[7]are also reported. Gravi- metric techniques are very accurate and can also be used for pure gas diffusion measurements[8,9]. The main dis- advantage of this type of method is the equilibration time, which can be as long as weeks, depending on the system.
The time to reach equilibrium can be minimized using a smaller quartz crystal microbalance (QCM)[10–15], which is the technique adopted in this work. The major advantage of this method is the short equilibration time, which ranges from 3 to 30 min, depending on the system, while retaining good accuracy.
QCM is based on the piezoelectric effect observed in an AT-cut quartz crystal. The quartz crystal wafer has golden electrode plates on both sides, creating a perpendicular elec- trical field between them. At fixed potential, the effect of the electrical field produces a stationary wave. The frequency of the wave is a function of the crystal mass so that a fre- quency change,F, can be related to a mass change,m, according to the Sauerbrey equation[10]:
F = −km (1)
where k is a proportionality constant that includes physical and geometrical properties of the crystal. Two quartz crys- tals are usually used: a working crystal coated with a thin layer of the polymer under study and an uncoated crystal is used as the reference. It is assumed that this coated mass is uniformly spread and vibrates synchronously with the quartz crystal. When gas, or solvent, is injected into the system the frequencies of both crystals change. This frequency change, FE, is a sum of three independent terms, hydrostaticFP, impedanceFV, and sorptionFS[16]:
FE=FS+FP+FV (2) FPis proportional to the pressure, whileFVis a function of the properties of the gas (viscosity and density) and of the crystal (base frequency, thickness and density). These two terms, FP and FV, were measured together using the reference crystal. For each temperature and pressure, the frequencies of the coated and the uncoated crystal were experimentally measured and their difference is proportional to mass of the coating, polymer and gas sorbed in it. Thus,
the solubility of the gas, C, in cm3gas (STP)/cm3polymer was calculated according to the following equation:
C= FSρpol
FCM ×22 416 (3)
whereρpol is the polymer density and M is the molecular weight of the gas and FC is the difference between the frequencies of the coated and the uncoated crystals after the equilibrium was reached.
3. Thermodynamic modeling
The wide variety of polymer applications, from paints and coatings to biotechnology and medical fields, clearly depicts the enormous amount of data that is in need. On the other hand, reliable methods to estimate polymer phase equilibrium could be used in the decision of the polymer application. In this work the Flory–Huggins model was used to correlate the experimental results. This model has been described extensively in the literature[17–21]and it has the advantage of being able to correlate experimental results of solubility of gases in polymers.
The Flory–Huggins model considers that the main con- tribution for the activity coefficient comes from the entropic effects rather than enthalpic effects. The entropy of mixing is calculated from the lattice theory and it accounts for the number of ways of arranging polymer and solvent molecules in a lattice, where each molecule occupies a number of sites proportional to its molecular volume. On the other hand, the enthalpy of mixing is calculated from the van Laar ener- getic term with the binary interaction parameter,χ12, which is used to provide an accurate representation of the experi- mental data
lnγ1=ln φ1
w1
+
1−1
r
φ2+χ12φ22 with r= v2M2
v1M1
(4) φ1= v1w1
v1w1+v2w2
and φ2=1−φ1 (5)
where γ1 is the weight fraction activity coefficient of the solvent,wi are the weight fractions of each i component, r is the number of segments in the polymer molecule, each segment having the same size as that of the solvent molecule, M1 is the solvent molecular weight and M2 is the weight average molecular weight of the polymer. The activity of the solvent a1, in the polymer is calculated by
a1=γ1w1 (6) FromEq. (1), the weight fraction of the solvent in polymer is given by
w1= FS
FS+FC
(7)
Table 1
Characteristicsaof polymers
Polymer Tg (K) Mw
Polystyrene 376.2 197000
Poly(lactic acid) 80:20 329.9 102800
a Tg =glass transition temperature, Mw =weight average molecular weight.
4. Materials and characterization methods
Atactic polystyrene was purchased from Polyscience, Inc.
and poly(lactic acid) was provided by Cargill–Dow polymers and their characteristics are shown inTable 1.
Carbon dioxide with 99.999% mol/mol minimum stated purity was purchased from Air Liquide. Nitrogen with 99.999% mol/mol minimum purity was used and it was obtained from Air Liquide. Dichloromethane was obtained from Riedel-deHaan with analytical reagent grade. Solvent and gases were used with no further purification.
A differential scanning calorimeter, DSC-50/DTA-50 from Shimadzu, was used to measure the glass transition temperature of PLA. A scanning electronic microscope, model S-4100 from Hitachi was used to study the polymer film surface with annealing time. The polymer surface to- pography was studied by atomic force microscopy, using a Multimode AFM with nanoscope IIIA controller from Digital Instruments.
The quartz crystals used were of 9 MHz base frequency, with golden electrodes of 3 mm and were supplied by Eu- roquartz, UK.
5. Apparatus description
The scheme of the QCM apparatus used in this work is presented inFig. 1. This apparatus operates in the pressure range from 2×10−6 to 1 bar and in the temperature range from 263 to 340 K. It was designed in a simple and versa- tile way, allowing the replacement of parts, according to the system under study. The solubility cell, which has a volume of 100 cm3 approximately, as well as the gas/vapor ther- mostat, which has a volume of 500 cm3 approximately, is placed inside of a large water bath, capable of maintaining the temperature to within 0.1 K. Two AT-cut quartz crys- tal of 9 MHz, the reference and the working crystals, were mounted in the electrical oscillator circuit, inside the solu- bility cell. The oscillator is powered by a stabilized potential of 5.0 V and both the quartz crystal frequencies were mea- sured using a 10 digits frequency counter HP 53131A, from Hewlett Packard. The temperature inside of the solubility cell is controlled with a previously calibrated Pt100 ther- mometer connected to a digital multimeter HP 974A from Hewlett Packard. A rotatory high vacuum pump RV3 from BOC Edwards, capable of reaching a vacuum of 0.08 mbar
T1 2
3
1 V2
99 V1
V4
11 V3
T
+/- 1 Hz
5,0 V 8
V5
6
P
7 5
10 12
4 13
Fig. 1. The QCM apparatus: 1, quartz crystals; 2, solubility cell; 3, oscillator; 4, gas/vapor thermostat; 5, power supply; 6, computer; 7, pressure transducer; 8, vacuum pump; 9, gas bottle; 10, frequency counter;
11, nitrogen trap; 12, multimeter; 13, Pt100 thermometer; V1–V5, valves;
T1, water bath.
was used while the system pressure was measured with an AEP transducer Bit02 of 4 digits.
The experiment starts by opening the valves V5, V3, V2 and V1 and the entire apparatus is kept in vacuum for 2 days.
After a given temperature is attained, valve V2 is closed and V1 is opened, inserting the gas from the gas bottle in- side the gas/vapor thermostat. A computer program controls the frequency counter and the transducer and records the frequency of the crystals and the pressure of the system. By opening valve V2 and closing valve V3, the gas is inserted into the solubility cell, promoting a change in the frequency, which is recorded as a function of time. When a stable fre- quency is reached, within an acceptable tolerance (±2 Hz), the reading is accepted and recorded as the equilibrium fre- quency. By closing valve V2 and opening valve V3, the gas in the solubility cell is evacuated and desorption process takes place.
6. Quartz crystal coating
The 9 MHz crystals were thoroughly cleaned with dichloromethane until the base frequency became constant (±2 Hz). Both sides of the working crystal were then coated with a polymer solution of known concentration and the solvent was left to evaporate under ambient conditions.
The quartz crystal coated with polymer was placed in an oven, 5◦ above its glass transition temperature for 2 days.
This temperature treatment relaxed the polymer chains, pro- ducing a more uniform coating, and further dried it, elim- inating contaminants. Cycles of temperature coupled with sorption measurements were performed until the sorption remained constant.
Fig. 2. SEM of the PLA sample coated on cover-glass with no temperature treatment.
In order to understand the effect of the temperature treatment on the polymer coating, scanning electronic mi- croscopy (SEM) was performed on several PLA coatings on glass. The sequence ofFigs. 2–4shows SEM images of the polymer film with different annealing time periods, 0, 48 and 109 h, respectively, at 333 K in an air-vented oven.
Fig. 5shows the SEM image of the polymer film when a cyclic annealing (38 h at 333 K, 105 h at ambient tempera- ture and 24 h more at 333 K) is done. These figures clearly show a more regular film (without cracks) in the last case and therefore this was the procedure adopted for preparing the quartz crystal coating during this work.
A more thorough study of the surface topography of the polymer on the crystal was performed using AFM, in the tapping mode[22].Fig. 6shows two dimension topographic images for a PLA coating that has been cyclically annealed, with a resolution of 100m. It can be observed that, at this magnitude, the surface is still rather irregular, with a max-
Fig. 3. SEM of the PLA sample coated on cover-glass that has been annealed at 333 K in air for 48 h.
Fig. 4. SEM of the PLA sample coated on cover-glass that has been annealed at 333 K in air for 109 h.
imum topographical variation (height) of less than 1.2m and an average roughness of 0.003m. The thickness of the film was also measured and has an estimated value of 2.6m.
7. Results
In order to estimate the precision of the QCM, the sorp- tion of carbon dioxide in polystyrene (PS) at pressures be- low the atmospheric pressure and at the temperatures of 303.2–313.2 K was measured. The solubility results listed in Table 2, have a precision of±0.02 cm3(STP) cm−3polymer and are compared with literature data[23–25]inFig. 7. It can be observed that a good description is obtained for both tem- peratures, with an overall average absolute deviation (AAD) of 3%. At 313.2 K the literature data depends strongly on the
Fig. 5. SEM of the PLA sample coated on cover-glass that has been annealed cyclically 38 h at 333 K 105 h at ambient temperature and 24 h at 333 K.
Fig. 6. Roughness analysis of PLA film coated on quartz crystal by AFM tapping force mode.
polymer weight average molecular weight (Mw) and orien- tation (atactic/syndiotactic). The polymer Mwcan be related to polymer density. For the same orientation, the sorption de- creases dramatically as the polymer density increases[26].
Note that the polymer orientation does not change the nature of the sorption process; however, it changes its magnitude.
For example, a decrease in the solubility can be correlated with the increase of density, due to orientation[24]. In this work, the PS weight average molecular weight is 197 000 and thus it agrees well with the literature results for similar weight average molecular weight and orientation.
The sorption isotherms obtained for carbon dioxide, oxy- gen, nitrogen and water in PLA at different pressures are dis- played inTables 3–6andFigs. 8–11, respectively. Analysis of the figures shows that water is the more soluble sorbent, even though the measurements could only be performed up to 90% of the water vapor pressure at that temperature. The atmospheric gases were measured up to atmospheric pres- sure and it was found that nitrogen and oxygen have ap-
Table 2
Solubility of carbon dioxide in PS
T (K) P (bar) C (cm3(STP) cm−3polymer)
303.2 0.214 0.46
0.510 1.04
0.767 1.53
0.824 1.63
0.983 1.95
313.2 0.361 0.72
0.711 1.30
0.769 1.42
0.968 1.73
0.0 0.5 1.0 1.5 2.0
0.0 0.2 0.4 0.6 0.8 1.0
P/bar C/cm3 (STP)cm-3 polymer
QCM at 303.2 K QCM at 313.2 K [24] at 303.2 K [24] at 313.2 K [25] at 313.2 K [25] at 313.2 K
Fig. 7. Solubility of carbon dioxide in PS and comparison with literature [24,25]. The PS Mw used in[24]was 290 000 and no information was given in [25]. The dashed line and the solid lines represent biaxially oriented and unoriented PS, respectively, at 313.2 K.
proximate solubility values, which are half of that found for carbon dioxide. Since there are no results available in the literature for these systems, no comparisons could be made.
InFigs. 8–11, the correlation of experimental results for each system at each temperature using the Flory–Huggins
Table 3
Solubility of carbon dioxide in PLA
T (K) P (bar) C (cm3(STP) cm−3polymer)
293.2 0.142 0.78
0.230 1.19
0.361 1.63
0.462 2.05
0.578 2.23
0.656 2.58
0.818 2.93
0.893 3.20
0.980 3.49
303.2 0.226 0.98
0.512 1.65
0.677 1.94
0.805 2.28
0.857 2.39
0.960 2.63
0.961 2.66
313.2 0.197 0.46
0.271 0.74
0.334 0.85
0.359 0.93
0.487 1.13
0.524 1.23
0.634 1.50
0.719 1.59
0.973 1.84
Table 4
Solubility of oxygen in PLA
T (K) P (bar) C (cm3(STP) cm−3polymer)
293.2 0.110 0.50
0.188 0.62
0.188 0.58
0.312 0.72
0.454 0.88
0.468 0.89
0.610 1.04
0.765 1.25
0.925 1.48
0.995 1.57
303.2 0.103 0.19
0.106 0.20
0.214 0.27
0.351 0.38
0.352 0.32
0.396 0.44
0.473 0.49
0.663 0.76
0.889 1.04
0.987 1.05
313.2 0.030 0.06
0.063 0.10
0.098 0.12
0.184 0.21
0.270 0.23
0.337 0.30
0.404 0.26
0.488 0.43
0.547 0.39
0.643 0.51
0.711 0.50
0.793 0.55
0.799 0.54
0.944 0.58
0.955 0.66
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
0.0 0.2 0.4 0.6 0.8 1.0
P/bar C/cm3 (STP)cm-3 polymer
QCM at 293.2 K QCM at 303.2 K QCM at 313.2 K FH at 293.2 K FH at 303.2 K FH at 313.2 K
Fig. 8. Solubility of carbon dioxide in PLA and the correlation with Flory–Huggins model.
Table 5
Solubility of nitrogen in PLA
T (K) P (bar) C (cm3(STP) cm−3polymer)
283.1 0.107 0.15
0.138 0.24
0.322 0.39
0.591 0.86
0.685 0.99
0.863 1.18
1.022 1.32
1.258 1.54
293.0 0.116 0.08
0.302 0.21
0.478 0.32
0.530 0.32
0.755 0.46
0.912 0.55
1.046 0.68
303.1 0.077 0.06
0.173 0.11
0.302 0.14
0.398 0.16
0.501 0.20
0.696 0.245
0.904 0.32
1.026 0.36
313.1 0.190 0.05
0.273 0.06
0.413 0.10
0.605 0.14
0.717 0.17
0.810 0.19
1.022 0.23
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
0.0 0.2 0.4 0.6 0.8 1.0
P/bar C/cm3 (STP)cm-3 polymer
QCM at 293.2 K QCM at 303.2 K QCM at 313.2 K FH at 293.2 K FH at 303.2 K FH at 313.2 K
Fig. 9. Solubility of oxygen in PLA and the correlation with Flory–Huggins model.
Table 6
Solubility of water in PLA
T (K) P (bar) C (cm3(STP) cm−3polymer)
293.2 0.003 2.18
0.007 5.21
0.007 4.80
0.012 7.87
0.015 10.30
0.018 12.95
0.020 15.96
0.024 19.45
303.2 0.005 1.61
0.011 3.10
0.011 3.20
0.012 4.00
0.014 4.81
0.020 6.50
0.022 7.83
0.026 9.67
0.032 11.85
0.035 14.47
313.2 0.004 0.67
0.011 1.60
0.017 3.33
0.026 5.25
0.035 7.03
0.043 9.81
0.043 9.73
0.045 9.45
0.046 10.20
model is depicted. The Flory–Huggins interaction parame- ters, listed inTable 7, were correlated for each temperature, assuming no concentration dependency. The analysis of the AAD shows that the deviations are larger for carbon diox- ide and oxygen than for nitrogen and water. The figures also confirm that the experimental data obtained for the first two
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
0.0 0.2 0.4 0.6 0.8 1.0
P/bar C/cm3 (STP)cm-3 polymer
283.1 K 293.0 K 303.1 K 313.1 K FH at 283.1 K FH at 293.0 K FH at 303.1 K FH at 313.1 K
Fig. 10. Solubility of nitrogen in PLA and the correlation with Flory–Huggins model.
0 5 10 15 20
0.00 0.01 0.02 0.03 0.04 0.05
P/bar C/cm3 (STP)cm-3 polymer
QCM at293.2 K QCM at 303.2 K QCM at 313.2 K FH at 293.2 K FH at 303.2 K FH at 313.2 K
Fig. 11. Solubility of water vapor in PLA and the correlation with Flory–Huggins model.
Table 7
Flory–Huggins parameter for gases in PLA and respective AAD
Solute T (K) χ12 AAD (%)
CO2 293.2 0.06 2
303.2 0.09 3
313.2 0.11 5
O2 293.2 1.40 2
303.2 1.76 14
313.2 2.10 8
N2 283.1 2.11 4
293.0 2.90 3
303.1 3.31 9
313.1 3.84 4
H2O 293.2 3.52 8
303.2 3.65 3
313.2 3.67 8
sorbents is less precise than for the last two. This fact is due to the successive upgrades performed in the QCM during the course of this work, namely improvements in the tem- perature and pressure control.
For the carbon dioxide/PLA system, the Flory–Huggins interaction parameter values are one order of magnitude smaller than for other systems. From a qualitative point of view, theχ12 is a measure of the solvent/polymer affinity, where low values ofχ12 describe the highest sorption, and thus the highest affinity.
8. Conclusions
The solubility carbon dioxide, oxygen, nitrogen and wa- ter in PLA at 293.2, 303.2 and 313.2 K, and pressures up to atmospheric pressure, were measured with a QCM. The results obtained during this work show that it is very sensi-
tive method, useful method to study gas–polymer sorption.
Conventional methods such as gravimetric or pressure de- cay have less sensitivity, require relatively larger samples and longer equilibrium times, usually hours or even days, depending on the system. QCM proved to be a reliable, pre- cise, accurate and fast method for studying the solubility of gases in polymers. The experimental results were correlated using Flory–Huggins model.
Acknowledgements
This project was financed by Fundação para a Ciˆencia e Tecnologia, POCTI/EQU/43356/2001. N.S. Oliveira thanks Fundação para a Ciˆencia e a Tecnologia the Ph.D. scholar- ship (SFRH/BD/6690/2001).
References
[1] G. Piringer, A.L. Baner, Plastic Packaging Materials for Food: Bar- rier Function, Mass Transport, Quality Assurance, and Legislation, Wiley/VCH, Weinheim, Germany, 2000.
[2] Z. Zhou, J. Springer, J. Appl. Polym. Sci. 47 (1993) 7.
[3] W.J. Koros, Sorption and Transport of Gases in Glassy Polymers, The University of Texas at Austin, Austin, USA, 1977.
[4] E. Morales, J.L. Acosta, Polym. J. 28 (1996) 127.
[5] R. Gavara, R. Catalã, S. Aucejo, D. Cabedo, R. Hernandez, Polymer 34 (1996) 1907.
[6] H.J. Lechermeier, J.R. Dorgan, J.D. Way, J. Membr. Sci. 4922 (2001) 1.
[7] F. Mantovani, M. Grassi, I. Colombo, R. Lapasin, Fluid Phase Equi- libr. 167 (2000) 63.
[8] J. Crank, The Mathematics of Diffusion, 2nd ed., Clarendon Press, Oxford, 1975.
[9] P. Hernandez-Muñoz, R. Gavara, J. Hernandez, J. Membr. Sci. 154 (1999) 195.
[10] G. Sauerbrey, Z. Phys. 155 (1959) 206D–222D.
[11] W.H. King, Anal. Chem. 36 (1964) 1735.
[12] H.C. Wong, S.W. Campbell, V.R. Bhethanabotla, Fluid Phase Equi- libr. 139 (1997) 371.
[13] M.T.S.R. Gomes, Desenvolvimento de Sensores Baseados em Mi- crobalanças de Cristais Piezoeléctricos para a Determinação de CO2
e SO2, Universidade de Aveiro, Aveiro, Portugal, 1996.
[14] S.P.V.N. Mikkilineni, D. Alan Tree, M.S. High, J. Chem. Eng. Data 40 (1995) 750.
[15] S. Saeki, J.C. Holste, D.C. Bonner, J. Polym. Sci., Polym. Phys. Ed.
19 (1981) 307.
[16] V.M. Mecea, Sens. Actuat. A 40 (1994) 1.
[17] K.S. Birdi, Handbook of Colloid and Surface Chemistry, 2nd ed., CRC Press, 2002.
[18] M. Kontogeorgis, I. Nikolopoulos, A. Fredenslund, P. Tassios, Fluid Phase Equilibr. 127 (1997) 103.
[19] H.C. Wong, S.W. Campbell, V.R. Bhethanabotla, Fluid Phase Equi- libr. 179 (2001) 181.
[20] J.M. Prausnitz, R.N. Lichtenthaler, E.G. de Azevedo, Molecular Ther- modynamics of Fluid-Phase Equilibria, Prentice-Hall PTR, NJ, USA, 1999.
[21] P.J. Flory, Statistical Thermodynamics of Polymer Solutions, Prin- ciples of Polymer Chemistry, Cornell University Press, Ithaca, NY, 1953 (Chapter XII).
[22] D. Raghavan, X. Gu, T. Nguyen, M. VanLandingham, A. Karim, Macromolecules 33 (2000) 2573.
[23] R. Paterson, Y. Yampol’skii, J. Phys. Chem. Ref. Data 28 (1999) 1264–1449.
[24] C. Carfagna, L. Nicodemo, L. Nicolais, G. Campanile, J. Polym.
Sci. Part B: Polym. Phys. 24 (1986) 1805.
[25] E. Sada, H. Kumazawa, H. Yakushji, Y. Bamba, K. Sakata, S.-T.
Wang, Ind. Eng. Chem. Res. 26 (1987) 433.
[26] S.C. George, S. Thomas, Prog. Polym. Sci. 26 (2001) 985.
